R/mcdina_est_item.R

In alexanderrobitzsch/CDM: Cognitive Diagnosis Modeling

## File Name: mcdina_est_item.R
## File Version: 0.19

#######################################################
# mcdina estimate item parameters
mcdina_est_item <- function( n.ik, lr_list, lc_list, delta, I, G,
        eps, itemstat, itempars, lr_counts )
{
    for (ii in 1:I){
        lr.ii <- lr_list[[ii]]
        lr_index.ii <- lr.ii[, "lr_index"]
        lc.ii <- lc_list[[ii]]
        #******************
        # group specific item parameters
        if ( itempars[ii]=="gr"){
            for (gg in 1:G){ # gg <- 1
                n.ik.ii.gg <- t( n.ik[ii,,,gg] )
                n.ik.ii.gg_aggr <- as.matrix( rowsum( n.ik.ii.gg, lr_index.ii   ) )
                n.ik.ii.gg_aggr1 <- n.ik.ii.gg_aggr+eps
                cn <- rowSums( n.ik.ii.gg_aggr1 )
                lr_counts[ii, 1:itemstat$N.lr[ii],  gg ] <- cn
                delta.new <- t( n.ik.ii.gg_aggr /
                        matrix( cn, nrow=nrow(n.ik.ii.gg_aggr),
                        ncol=ncol(n.ik.ii.gg_aggr), byrow=FALSE ) )
                delta[ii,,1:itemstat[ii,"N.lr"],gg] <- delta.new
            }
        }
        #******************
        # group specific item parameters
        if ( itempars[ii] !="gr"){
            n.ik.ii.gg <- t( n.ik[ii,,,1] )
            for (gg in 2:G){
                n.ik.ii.gg <- n.ik.ii.gg + t( n.ik[ii,,,1] )
            }
            n.ik.ii.gg_aggr <- as.matrix( rowsum( n.ik.ii.gg, lr_index.ii   ) )
            n.ik.ii.gg_aggr1 <- n.ik.ii.gg_aggr+eps
            cn <- rowSums( n.ik.ii.gg_aggr1 )
            lr_counts[ii, 1:itemstat$N.lr[ii],  1:G ] <- cn
            delta.new <- t( n.ik.ii.gg_aggr /
                            matrix( cn, nrow=nrow(n.ik.ii.gg_aggr),
                        ncol=ncol(n.ik.ii.gg_aggr), byrow=FALSE ) )
            for (gg in 1:G){
                delta[ii,,1:itemstat[ii,"N.lr"],gg] <- delta.new
            }
        }  # end itempars[ii] !="gr"
    }  # end ii
    res <- list("delta"=delta, "lr_counts"=lr_counts )
    return(res)
}
############################################################